Mumax 2016

From HPC users
Jump to navigationJump to search


mumax3 is a GPU-accelerated micromagnetic simulation program developed and maintained at the DyNaMat group at Ghent University.

A speed-up of the order of 100x compared to CPU-based simulations can easily be reached, even with relatively inexpensive gaming GPUs. Additionally, the software is optimized for low memory use and can handle about 16 million FD cells with 2GB of GPU RAM. 1

Installed version(s)

The following versions are installed and currently available...

... on environment hpc-env/8.3:

  • mumax/3.10-GCCcore-8.3.0-CUDA-11.4.2

Loading mumax

To load the desired version of the module, use the module load command, e.g.

module load hpc-env/8.3
module load mumax

Always remember: this commands are case sensitive!

Using mumax

To find out on how to use mumax you can just type in mumax3 --help after loading the module to print out a help text to get you started:

$ mumax3 --help
Usage of mumax3:
  -cache string
        Kernel cache directory (empty disables caching) (default "/scratch/24403396_erle1100")
  -f    Force start, clean existing output directory
        If one simulation fails, stop entire batch immediately
  -gpu int
        Specify GPU
  -http string
        Port to serve web gui (default ":35367")
  -i    Open interactive browser session
  -o string
        Override output directory
        Enable convolution self-test for cuFFT sanity.
  -s    Silent
        Synchronize all CUDA calls (debug)
        Cuda test (internal)
  -v    Print version (default true)
        Check input files for errors, but don't run them

Additionally, there are two more binaries for you to use after loading the mumax module: mumax3-convert and mumax3-server

$ mumax3-convert --help
Usage of mumax3-convert:
  -arrows int
        Arrow size for vector bitmap image output
  -color string
        Colormap for scalar image output. (default "black,gray,white")
  -comp string
        Select a component of vector data. (0,1,2 or x,y,z)
        output in CSV format
        output in dump format
  -f string
        Printf format string (default "%v")
        GIF output
        Gnuplot-compatible output
        JPEG output
        output in JSON format
  -max string
        Maximum of color scale: "auto" or value. (default "auto")
  -min string
        Minimum of color scale: "auto" or value. (default "auto")
        Normalize vector data to unit length
        Scale vector data, maximum to unit length
        Numpy output
  -o string
        Save all output in this directory
  -omf string
        "text" or "binary" OVF1 output
  -ovf string
        "text" or "binary" OVF1 output
  -ovf2 string
        "text" or "binary" OVF2 output
        PNG output
  -resize string
        Resize. E.g.: 128x128x4
        Human-readible output to stdout
        SVG output
        SVGZ output (compressed)
  -vtk string
        "ascii" or "binary" VTK output
  -xrange string
        Crop x range min:max (both optional, max=exclusive)
  -yrange string
        Crop y range min:max (both optional, max=exclusive)
  -zrange string
        Crop z range min:max (both optional, max=exclusive)

$ mumax3-server --help
Usage of mumax3-server:
  -cache string
        mumax3 kernel cache path
  -exec string
        mumax3 executable (default "mumax3")
  -halflife duration
        share decay half-life (default 24h0m0s)
  -l string
        Listen and serve at this network address (default ":35360")
        log debug output (default true)
  -ports string
        Scan these ports for other servers (default "35360-35361")
  -scan string
        Scan these IP address for other servers (default "")
  -timeout duration
        Portscan timeout (default 2s)


The full documentation can be found here.